UPDATED 02.13.2023

wordhoard/utilities/basic_soup.py

@@ -23,7 +23,7 @@
 # Date Completed: October 15, 2020
 # Author: John Bumgarner
 #
-# Date Last Revised: February 04, 2023
+# Date Last Revised: February 10, 2023
 # Revised by: John Bumgarner
 ##################################################################################
 

wordhoard/utilities/captcha_checker.py

@@ -0,0 +1,115 @@
+#!/usr/bin/env python3
+
+"""
+This Python script is used to verify the HREF being queried is
+protected by a captcha that requires human interaction to bypass.
+"""
+__author__ = 'John Bumgarner'
+__date__ = 'December 07, 2021'
+__status__ = 'Production'
+__license__ = 'MIT'
+__copyright__ = 'Copyright (C) 2021 John Bumgarner'
+
+##################################################################################
+# “AS-IS” Clause
+#
+# Except as represented in this agreement, all work produced by Developer is
+# provided “AS IS”. Other than as provided in this agreement, Developer makes no
+# other warranties, express or implied, and hereby disclaims all implied warranties,
+# including any warranty of merchantability and warranty of fitness for a particular
+# purpose.
+##################################################################################
+
+##################################################################################
+# Date Completed: December 07, 2021
+# Author: John Bumgarner
+#
+# Date Revised:
+# Revised by:
+##################################################################################
+
+##################################################################################
+# Python imports required for basic operations
+##################################################################################
+import logging
+
+logger = logging.getLogger(__name__)
+
+
+class CaptchaVerification(object):
+    """
+    This Class is used to query a webpage to determine if it is protected by a captcha.
+    """
+
+    def __init__(self, url, soup):
+        self._url = url
+        self._raw_soup = soup
+
+    def _check_div_tag(self):
+        """
+        This function is designed to query for the existence
+        of specific tag known to be commonly related to a
+        captcha protected webpage.
+
+        :return: True or False
+        :rtype: boolean
+        """
+        captcha_protected_div_tag = bool(self._raw_soup.find(name='div', attrs={'id': 'px-captcha'}))
+        if captcha_protected_div_tag is True:
+            logger.error(f'The {self._url} has captcha protection.')
+            return True
+        elif captcha_protected_div_tag is False:
+            return False
+
+    def _check_title_tag(self):
+        """
+        This function is designed to query for the existence
+        of specific captcha related text contained in a title tag.
+
+        :return: True or False
+        :rtype: boolean
+        """
+        title_tag = self._raw_soup.find(name='title')
+        if 'Are you a robot?' in title_tag.text:
+            logger.error(f'The {self._url} has captcha protection.')
+            return True
+        else:
+            return False
+
+    def _check_h2_tag(self):
+        """
+        This function is designed to query for the existence
+        of specific captcha related text contained in an H2 tag.
+
+        :return: True or False
+        :rtype: boolean
+        """
+        captcha_protected_h2_tag = bool(self._raw_soup.find(name='h2', attrs={'class': 'main__heading'}))
+        if captcha_protected_h2_tag is True:
+            captcha_protected_tag = self._raw_soup.find(name='h2', attrs={'class': 'main__heading'})
+            if "We've detected unusual activity from your computer network" in captcha_protected_tag.text:
+                return True
+            else:
+                return False
+
+    def captcha_protected_url(self):
+        """
+        This function is designed to query specific elements, which
+        will determine if a webpage is protected by a captcha.
+
+        :return: True or False
+        :rtype: boolean
+        """
+        div_tag_bool = self._check_div_tag()
+        if div_tag_bool is True:
+            return True
+        elif div_tag_bool is False:
+            title_tag_bool = self._check_title_tag()
+            if title_tag_bool is True:
+                return True
+            else:
+                h2_tag_bool = self._check_h2_tag()
+                if h2_tag_bool is True:
+                    return True
+                else:
+                    return False

wordhoard/utilities/colorized_text.py

@@ -1,7 +1,7 @@
 #!/usr/bin/env python3
 
 """
-This Python script provide colorized text in error messages thrown by WordHoard.
+This Python script provide colorized text in the error messages thrown by WordHoard.
 """
 __author__ = 'John Bumgarner'
 __date__ = 'February 04, 2023'

wordhoard/utilities/deep_translator.py

@@ -2,7 +2,7 @@
 
 """
 This Python script is used to translate a specific word from it source language,
-such as Spanish into American English using the Deep translation service.
+such as Spanish into American English using the Deep Translation service.
 """
 __author__ = 'John Bumgarner'
 __date__ = 'September 24, 2021'
@@ -266,6 +266,8 @@ def translate_word(self):
             print(colorized_text(255, 0, 0, f'The language provided is not one of the supported languages '
                                             f'for the Deep Translation service.'))
             print(colorized_text(255, 0, 0, f'Requested language: {self._source_language}'))
+            print(colorized_text(255, 0, 0, f'Please review the languages supported by the Deep Translate service\n'
+                                            f'https://wordhoard.readthedocs.io/en/latest/translations/deepl_supported_translation_languages/'))
             return None
 
     def reverse_translate(self):

wordhoard/utilities/email_address_verification.py

@@ -29,7 +29,6 @@
 #
 ##################################################################################
 
-
 ##################################################################################
 # Python imports required for basic operations
 ##################################################################################

wordhoard/utilities/exceptions.py

@@ -1,7 +1,7 @@
 #!/usr/bin/env python3
 
 """
-This Python script provides various Exceptions, which are used in the WordHoard Translation Classes.
+This Python script provides various Exceptions Classes, which are used in the WordHoard Translation modules.
 """
 __author__ = 'John Bumgarner'
 __date__ = 'February 04, 2023'

wordhoard/utilities/google_translator.py

@@ -2,7 +2,7 @@
 
 """
 This Python script is used to translate a specific word from it source language,
-such as Spanish into American English using the Google translation service.
+such as Spanish into American English using the Google Translation service.
 """
 __author__ = 'John Bumgarner'
 __date__ = 'September 24, 2021'
@@ -194,7 +194,7 @@ def _google_translate(self, original_language):
         except RequestException as error:
             """
             This exception is thrown when an ambiguous exception occurs during a connection to the 
-            Google Translation service. Translator service.
+            Google Translation service.
             """
             print(colorized_text(255, 0, 0, 'An ambiguous connection exception has occurred when contacting the'
                                             'Google Translation service.  Please check the WordHoard log file '
@@ -261,7 +261,7 @@ def _google_translate_reverse(self):
         except RequestException as error:
             """
             This exception is thrown when an ambiguous exception occurs during a connection to the 
-            Google Translation service. Translator service.
+            Google Translation service.
             """
             print(colorized_text(255, 0, 0, 'An ambiguous connection exception has occurred when contacting the'
                                             'Google Translation service.  Please check the WordHoard log file '
@@ -286,6 +286,8 @@ def translate_word(self):
             print(colorized_text(255, 0, 0, f'The language provided is not one of the supported languages '
                                             f'for the Google Translation service.'))
             print(colorized_text(255, 0, 0, f'Requested language: {self._source_language}'))
+            print(colorized_text(255, 0, 0, f'Please review the languages supported by the Google Translation service\n'
+                                            f'https://wordhoard.readthedocs.io/en/latest/translations/google_supported_translation_languages/'))
             return None
 
     def reverse_translate(self):

wordhoard/utilities/mymemory_translator.py

@@ -2,7 +2,7 @@
 
 """
 This Python script is used to translate a specific word from it source language,
-such as Spanish into American English using the MyMemory translation service.
+such as Spanish into American English using the MyMemory Translation service.
 """
 __author__ = 'John Bumgarner'
 __date__ = 'September 24, 2021'
@@ -116,7 +116,7 @@ def _mymemory_supported_languages(self):
              service.
             """
             logger.info(f'The language provided is not one of the supported languages of the MyMemory '
-                        f'translation service.')
+                        f'Translation service.')
             logger.info(f'Requested language: {self._source_language}')
             logger.error(''.join(traceback.format_tb(error.__traceback__)))
 
@@ -151,7 +151,7 @@ def _mymemory_translate(self, original_language):
         try:
             if validate_address(self._email_address) is False:
                 print(colorized_text(255, 0, 0,
-                                     'A valid email address is required to use the MyMemory translation '
+                                     'A valid email address is required to use the MyMemory Translation '
                                      'service.'))
                 sys.exit(1)
             elif validate_address(self._email_address) is True:
@@ -187,9 +187,9 @@ def _mymemory_translate(self, original_language):
             not the validity of the address provided.
             """
             print(colorized_text(255, 0, 0, 'The email address provided for authentication to the MyMemory '
-                                            'Translation MyMemory is invalid.'))
+                                            'Translation service is invalid.'))
             logger.error('Invalid Email Address Error:')
-            logger.error('The email address provided for authentication to the MyMemory Translation MyMemory '
+            logger.error('The email address provided for authentication to the MyMemory Translation service '
                          'is invalid.')
             logger.error(''.join(traceback.format_tb(error.__traceback__)))
 
@@ -198,7 +198,7 @@ def _mymemory_translate(self, original_language):
             This exception is thrown if the provided text exceed the length limit of the MyMemory Translator service.
             """
             logger.error(f'The text length for the word: {self._str_to_translate} exceed the length limit of '
-                         f'MyMemory translation service.')
+                         f'MyMemory Translation service.')
             logger.error(''.join(traceback.format_tb(error.__traceback__)))
 
         except TooManyRequestsException as error:
@@ -235,7 +235,7 @@ def _mymemory_translate_reverse(self):
         try:
             if validate_address(self._email_address) is False:
                 print(colorized_text(255, 0, 0,
-                                     'A valid email address is required to use the MyMemory translation '
+                                     'A valid email address is required to use the MyMemory Translation '
                                      'service.'))
                 sys.exit(1)
             elif validate_address(self._email_address) is True:
@@ -271,9 +271,9 @@ def _mymemory_translate_reverse(self):
             not the validity of the address provided.
             """
             print(colorized_text(255, 0, 0, 'The email address provided for authentication to the MyMemory '
-                                            'Translation MyMemory is invalid.'))
+                                            'Translation is invalid.'))
             logger.error('Invalid Email Address Error:')
-            logger.error('The email address provided for authentication to the MyMemory Translation MyMemory '
+            logger.error('The email address provided for authentication to the MyMemory Translation '
                          'is invalid.')
             logger.error(''.join(traceback.format_tb(error.__traceback__)))
             sys.exit(1)
@@ -307,7 +307,7 @@ def _mymemory_translate_reverse(self):
                                             'for additional information.'))
             logger.error('Connection Exception:')
             logger.error('An ambiguous connection exception has occurred when communicating with the '
-                         'MyMemoryTranslation service.')
+                         'MyMemory Translation service.')
             logger.error(''.join(traceback.format_tb(error.__traceback__)))
 
     def translate_word(self):
@@ -324,9 +324,8 @@ def translate_word(self):
             print(colorized_text(255, 0, 0, f'The language provided is not one of the supported languages '
                                             f'for the MyMemory Translation service.'))
             print(colorized_text(255, 0, 0, f'Requested language: {self._source_language}'))
-            print(colorized_text(255, 0, 0, f'Please review Mymemory'
-                                            f'https://wordhoard.readthedocs.io/en/latest/supported_languages'
-                                            f'for the Mymemory'))
+            print(colorized_text(255, 0, 0, f'Please review the languages supported by the MyMemory Translate service\n'
+                                            f'https://wordhoard.readthedocs.io/en/latest/translations/mymemory_supported_translation_languages/'))
 
             return None
 

wordhoard/utilities/translator_languages.py

@@ -1,8 +1,8 @@
 #!/usr/bin/env python3
 
 """
-This Python script provides the supported languages for the translation services used
-by WordHoard.
+This Python script provides the supported languages for the translation modules
+embedded into WordHoard.
 """
 __author__ = 'John Bumgarner'
 __date__ = 'February 12, 2023'
@@ -34,7 +34,7 @@ class Languages:
     @staticmethod
     def deep_supported_languages():
         """
-        This function returns the supported languages for the Deep translation service.
+        This function returns the supported languages for the Deep Translation service.
 
         :return: languages
         :rtype: string
@@ -75,7 +75,7 @@ def deep_supported_languages():
     @staticmethod
     def google_supported_languages():
         """
-        This function returns the supported languages for the Google Translator service.
+        This function returns the supported languages for the Google Translation service.
 
         :return: languages
         :rtype: string
@@ -221,7 +221,7 @@ def google_supported_languages():
     @staticmethod
     def mymemory_supported_languages():
         """
-        This function returns the supported languages for the MyMemory Translator services.
+        This function returns the supported languages for the MyMemory Translation service.
 
         :return: languages
         :rtype: string

wordhoard/utilities/word_verification.py

@@ -1,7 +1,8 @@
 #!/usr/bin/env python3
 
 """
-This Python script is to validate the syntax of words to query.
+This Python script is designed to validate the syntax of word in a query
+meets a format requirement.
 """
 __author__ = 'John Bumgarner'
 __date__ = 'October 15, 2020'